基于shearlet与PCNN的多聚焦图像融合方法

针对传统脉冲耦合神经网络(PCNN)模型在多聚焦图像融合领域应用中面临的参数繁杂等问题,提出一种基于剪切波(shearlet)变换与改进型PCNN的多聚焦图像融合方法。相比以往的变换域方法,shearlet具有理想的图像信息捕捉性能以及较低的计算复杂度,因此,可利用shearlet将待融合图像进行多尺度多方向分解。其次,对经典PCNN模型加以改进,综合运用清晰度水平以及协调矩阵完成低频子带图像以及一系列高频子带图像的融合过程。最后,运行shearlet反变换得到最终融合图像。仿真实验选取了若干组待融合图像进行仿真,验证了该方法在主、客观评价两方面的优越性。

Fusion Method for Multi-focus Images Based on Shearlet and Pulse Coupled Neural Networks

In order to settle the problem that there are so many parameters in the Pulse Coupled Neural Networks (PCNN)model when used in the area of multi-focus image fusion,a fusion method for multi-focus images based on shearlet transform and improved PCNN is proposed. Compared with the past transform domain methods,shearlet possesses much better ability to capture the image information and much lower computational complexities,so it can be used to conduct the multi-scale and multi-directional decompositions towards the source images. Besides,with the definition level and coordinate matrix,the classic PCNN model can be improved to fuse the low-frequency sub-images and a series of high-frequency ones. Finally,the final fused image can be obtained with the inverse shearlet. Several different groups of source images have been chosen in the simulation experiments,the proposed method proves to be superior in both subjective and objective aspects.